[Eeglablist] Epoch Length, and Normalization

[Tarik S Bel-Bahar]

Greetings Joseph, some thoughts below, cheers!



1. If there was a stimulus every 3000 ms or so in both Block1 and Block2
trials,
then you should be able to cut each Block 1 trial into one -500 to 3000ms
epoch
and for each Block 2  trial into about 17 epochs of -500 to 3000ms . So
you'll have more epochs from Block 2.
Having similar epochs across the Blocks is better for eeglab, and makes it
easier for you to compare the conditions later.
Using different length epochs if you have a good a-priori reason to from
the published literature.

2. You may want to google past eeglab posts on baselining. See also
mentions there of Groppe et al., paper on baselining or not before ICA. See
also two baselining paper titles below.

3. Try both, it won't hurt, and it should look similar. I think baselining
before ICA is recommended in Makoto's online pipeline list but please
double-check.

4.  Normalizing usually means that the units have been modified to be more
comparable across subjects (e.g., to -1 to 1). Depends on how you normalize
the signal in question, and whether you normalize by channel(s), time(s),
or subject.  Probably best to review descriptions of specific kinds of
normalization you'd like to emulate in the published literature, and go
from there.  Also,  take a look at the values in the channel power before
and after normalizing.


Single-Trial Normalization for Event-Related Spectral Decomposition
Reduces Sensitivity to Noisy Trials

Single-trial time–frequency analysis of electrocortical signals:
*Baseline *correction
and beyond





On Tue, Jun 28, 2016 at 3:09 PM, Joseph Nuamah <jknuamah at aggies.ncat.edu>
wrote:

> Dear Everyone,
>
> I have a data set from an experiment that had two blocks of trials.
>
> Each block had 30 trials. Whereas each trial in block 1 had a duration of
> 6000ms, each trial in block 2 had a duration of 60 000 ms. The ISI for both
> blocks was 3000 ms.
>
> The sampling frequency was 256Hz. A trial ended when a participant clicked
> the mouse button. I found out that for both cases, participants clicked the
> mouse before elapse of the stipulated trial duration.
>
> My questions are:
>
> 1. What epoch lengths do I use for both blocks? Are -500ms to 3000ms, and
> -500 to 5000ms okay for block 1 and block 2 respectively?
>
> 2. For time-frequency transforms, do I need to subtract the baseline from
> the data before I do the transformation?
>
> 3. As a follow up from question 2, do I perform ICA before the subtraction?
>
> 4. What is the effect of normalizing the data before computing the power
> in each channel?
>
> Thanks.
>
> Joseph.
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20160629/12b613e7/attachment.html>